A method for quickly marking arrows on road point clouds

Abstract

The present invention discloses a method for quickly marking arrows on a road point cloud, and relates to the technical field of road point clouds. The method for quickly marking arrows on a road point cloud comprises the following steps: S1, point cloud data collection; S2, marking key points on the point cloud; S3, calculating transformation relationships; S4, applying transformation parameters. The method for quickly marking arrows on a road point cloud only requires manually marking key points in an existing arrow template, and then performing a key point comparison based on the arrow template and the actual road arrow to obtain corresponding scaling, rotation, and translation transformation parameters, and applying the transformation to all points in the template to obtain the positions of all arrow points in the actual road arrow point cloud, thereby solving the problems of long time consumption, low efficiency, and low accuracy caused by manually setting arrow points, and is suitable for large-scale data set operations after the template is generated, and can improve accuracy and operating efficiency due to the reduction of manual operations.

Description

Technical Field

[0001] The present invention relates to the technical field of road surface point clouds, and in particular to a method for quickly marking arrows on road surface point clouds. Background Art

[0002] With the development of 3D laser scanning technology, point cloud data has become an important means of obtaining road information. Point cloud data can accurately reflect the three-dimensional shape of the road surface, providing a rich source of information for road planning, maintenance and traffic management. By annotating the elements in the road point cloud, it can provide true value data for autonomous driving algorithm training.

[0003] For example, the patent with application number: CN202311573498.6 discloses a road arrow accuracy assessment method, device, equipment and storage medium, including: obtaining road arrow feature points, extracting the original point cloud of road arrow feature points, obtaining the target point cloud, performing point cloud strength calculation processing on the target point cloud, generating the geometric confidence corresponding to the road arrow, performing logical calculation processing on the road arrow feature points, generating the logical confidence corresponding to the road arrow, fusing the geometric confidence and the logical confidence, and determining the road arrow accuracy assessment result. The present invention associates the accuracy of the road arrow element with the surrounding original point cloud, and automatically obtains the confidence of the road arrow by integrating the geometric accuracy and logical consistency of the road arrow, thereby generating a quantitative and accurate road arrow accuracy evaluation result, improving the accuracy of subsequent positioning and regulation and control, and meeting vehicle positioning requirements;

[0004] Applications similar to the above patents have the following defects:

[0005] Although some automated algorithms have been proposed for point cloud data processing, these methods often require complex preprocessing steps and have low accuracy when processing large-scale data sets.

[0006] Therefore, in view of this, the existing structure and deficiencies are studied and improved, and a method for quickly marking arrows on road point clouds is proposed. Summary of the invention

[0007] In view of the deficiencies of the prior art, the present invention provides a method for quickly marking arrows on a road surface point cloud, which solves the problems raised in the above-mentioned background technology.

[0008] To achieve the above objectives, the present invention is implemented by the following technical solution: a method for quickly marking arrows on a road point cloud, the method for quickly marking arrows on a road point cloud comprising the following steps:

[0009] S1. Point cloud data collection:

[0010] Scanning the actual road surface using a 3D laser scanning system to generate original point cloud data containing road surface information;

[0011] S2. Annotate key points on the point cloud:

[0012] Identify and extract key points of road arrows from the original point cloud data through the point cloud data annotation platform;

[0013] S3. Calculate the transformation relationship:

[0014] Compare the three key points in the original point cloud data generated by the actual road surface with the three key points corresponding to the template to obtain the corresponding scaling, rotation, and translation transformation parameters;

[0015] S4. Application of transformation parameters:

[0016] The key points in the template apply the transformation parameters to obtain the positions of all points including the key points indicated by arrows in the original point cloud data generated by the actual road surface.

[0017] Furthermore, in step S3, the transformation of parameters includes the following operations:

[0018] The three key points in the point cloud of the actual road arrow are A, B, and C, and the key points of the arrow template are A_t, B_t, and C_t;

[0019] Then calculate the scaling ratio and find the distance between any two of the three points:

[0020] Dis_AB,Dis_BC,Dist_AC,Dis_AB_t,Dis_BC_t,Dis_AC_t;

[0021] Then calculate the distance ratio:

[0022] ScaleRatio_AB=Dis_AB / Dis_AB_t,ScaleRatio_BC=Dis_BC / Dis_BC_t,ScaleR

[0023] atio_AC=Dis_AC / Dis_AC_t;

[0024] After obtaining the three distance ratios, calculate the average of the three distance ratios to obtain the same scaling ratio for all template key points:

[0025] ScaleRatio_avg=(ScaleRatio_AB+ScaleRatio_BC+ScaleRatio_AC) / 3;

[0026] Then all arrow template key points are scaled by the ratio ScaleRatio_avg.

[0027] Furthermore, the transformation parameters also include the following operations:

[0028] The three-dimensional Euclidean transformation matrix from the scaled three arrow template points to the three point cloud key points of the actual road arrow is calculated, and the three-dimensional Euclidean transformation matrix is ​​solved by singular value decomposition.

[0029] Furthermore, the transformation parameters also include the following operations:

[0030] The calculated three-dimensional Euclidean transformation matrix is ​​applied to all the scaled arrow template points to obtain arrow annotations that fit the three-dimensional point cloud of the road surface, thereby determining the size and direction information of the actual road surface arrows.

[0031] Furthermore, the arrow template is adapted to the actual road arrows.

[0032] Furthermore, in step S1, the road surface information refers to road surface elements, which include but are not limited to arrows and lane lines.

[0033] Furthermore, the method for quickly marking arrows on a road surface point cloud is applied to the field of road surface arrow marking.

[0034] The present invention provides a method for quickly marking arrows on a road surface point cloud, which has the following beneficial effects:

[0035] 1. The method for quickly marking arrows on a road point cloud only requires manually marking key points in all points of the point cloud data generated by an existing arrow template, and then performing a key point comparison based on the arrow template and the actual road arrow to obtain corresponding scaling, rotation, and translation transformation parameters, and applying the transformation to all points in the template to obtain the positions of all arrow points in the actual road arrow point cloud, thereby solving the problems of long time consumption, low efficiency, and low accuracy caused by manually setting arrow points. The method reduces the manual operation content and the adverse effects of manual marking on accuracy, and can be applied to large-scale data set operations after the template is generated, and the accuracy and operation efficiency can be improved due to the reduction of manual operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 The present invention is a schematic flow chart of the steps of a method for quickly marking arrows on a road surface point cloud. DETAILED DESCRIPTION

[0037] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0038] like Figure 1As shown, the present invention provides a technical solution: a method for quickly marking arrows on a road surface point cloud, and the method for quickly marking arrows on a road surface point cloud includes the following steps:

[0039] S1. Point cloud data collection:

[0040] Scanning the actual road surface using a 3D laser scanning system to generate original point cloud data containing road surface information;

[0041] By scanning the actual road surface with a 3D laser scanning system, various road surface elements can be obtained, including arrows, lane lines, etc.

[0042] S2. Annotate key points on the point cloud:

[0043] Identify and extract key points of road arrows from the original point cloud data through the point cloud data annotation platform;

[0044] S3. Calculate the transformation relationship:

[0045] Compare the three key points in the original point cloud data generated by the actual road surface with the three key points corresponding to the template to obtain the corresponding scaling, rotation, and translation transformation parameters;

[0046] When comparing the three key points in the original point cloud data generated by the actual road surface with the three key points corresponding to the template, it is necessary to first select the corresponding arrow template according to the arrow shape extracted from the actual road surface;

[0047] Transforming parameters involves the following operations:

[0048] The three key points in the point cloud of the actual road arrow are A, B, and C, and the key points of the arrow template are A_t, B_t, and C_t, where the arrow template is adapted to the actual road arrow;

[0049] Then calculate the scaling ratio and find the distance between any two of the three points:

[0050] Dis_AB,Dis_BC,Dist_AC,Dis_AB_t,Dis_BC_t,Dis_AC_t;

[0051] Then calculate the distance ratio:

[0052] ScaleRatio_AB=Dis_AB / Dis_AB_t,ScaleRatio_BC=Dis_BC / Dis_BC_t,ScaleR

[0053] atio_AC=Dis_AC / Dis_AC_t;

[0054] After obtaining the three distance ratios, calculate the average of the three distance ratios to obtain the same scaling ratio for all template key points:

[0055] ScaleRatio_avg=(ScaleRatio_AB+ScaleRatio_BC+ScaleRatio_AC) / 3;

[0056] Then all arrow template key points are scaled by the scale ratio ScaleRatio_avg;

[0057] Then calculate the 3D Euclidean transformation matrix from the scaled 3 arrow template points to the 3 point cloud key points of the actual road arrows. They are three sets of non-collinear points that correspond one to one, so they can be solved in the same way as the common ICP registration of two sets of corresponding point clouds (refer to Gao Xiang's "Fourteen Lectures on Visual Slam"), and the 3D Euclidean transformation matrix can be solved by singular value decomposition (SVD);

[0058] Apply the calculated three-dimensional Euclidean transformation matrix to all the arrow template points after scaling, thereby obtaining arrow annotations that fit the three-dimensional point cloud of the road surface, thereby determining the size and direction information of the actual road surface arrows;

[0059] S4. Application of transformation parameters:

[0060] The transformation parameters are applied to the key points in the template to obtain the positions of all points including the key points indicated by arrows in the original point cloud data generated by the actual road surface.

[0061] The method for quickly marking arrows on a road surface point cloud is applied to the field of road surface arrow marking. It only requires manual marking of key points in all points of the point cloud data generated by an existing arrow template, and then performing key point comparison based on the arrow template and the actual road surface arrow to obtain corresponding scaling, rotation, and translation transformation parameters, and then applying the transformation to all points in the template to obtain the positions of all arrow points in the actual road surface arrow point cloud, thereby solving the problems of long time consumption, low efficiency, and low accuracy caused by manual setting of arrow points. Thus, by reducing the manual operation content, the adverse effects of manual marking on accuracy are reduced, and after the template is generated, it can be applied to large-scale data set operations, and the accuracy and operation efficiency can be improved due to the reduction of manual operations.

[0062] In summary, the method for quickly marking arrows on a road surface point cloud is, when used, firstly, a template is made based on various arrows on the road surface, firstly, original point cloud data containing road surface information is generated, and three key points that can best represent the information contained in the arrow are selected from the original point cloud data of each template, and each template and its three key points are associated and stored in a point cloud data storage platform; the selection position of the key point is the side corner position of the road surface arrow, and the key point represents the information contained in the arrow as follows: based on the position of the key point, the arrow size information and the direction information represented by the arrow can be included, and the original point cloud data of each template also includes the positions of the remaining points except the key point;

[0063] The actual road surface is scanned by a 3D laser scanning system to generate original point cloud data containing road surface information; various road surface elements, including arrows, lane lines, etc., can be obtained by scanning the actual road surface with a 3D laser scanning system;

[0064] Identify and extract key points of road arrows from the original point cloud data through the point cloud data annotation platform;

[0065] The three key points in the original point cloud data generated by the actual road surface are compared with the three key points corresponding to the template to obtain the corresponding scaling, rotation and translation transformation parameters, and the key points in the template are applied with the transformation parameters to obtain the positions of all points including the key points in the arrow in the original point cloud data generated by the actual road surface; when the three key points in the original point cloud data generated by the actual road surface are compared with the three key points corresponding to the template, the corresponding arrow template needs to be selected according to the arrow shape extracted from the actual road surface;

[0066] Transforming parameters involves the following operations:

[0067] The three key points in the point cloud of the actual road arrow are A, B, and C, and the key points of the arrow template are A_t, B_t, and C_t, where the arrow template is adapted to the actual road arrow;

[0068] Then calculate the scaling ratio and find the distance between any two of the three points:

[0069] Dis_AB,Dis_BC,Dist_AC,Dis_AB_t,Dis_BC_t,Dis_AC_t;

[0070] Then calculate the distance ratio:

[0071] ScaleRatio_AB=Dis_AB / Dis_AB_t,ScaleRatio_BC=Dis_BC / Dis_BC_t,ScaleR

[0072] atio_AC=Dis_AC / Dis_AC_t;

[0073] After obtaining the three distance ratios, calculate the average of the three distance ratios to obtain the same scaling ratio for all template key points:

[0074] ScaleRatio_avg=(ScaleRatio_AB+ScaleRatio_BC+ScaleRatio_AC) / 3;

[0075] Then all arrow template key points are scaled by the scale ratio ScaleRatio_avg;

[0076] Then calculate the 3D Euclidean transformation matrix from the scaled 3 arrow template points to the 3 point cloud key points of the actual road arrows. They are three sets of non-collinear points that correspond one to one, so they can be solved in the same way as the common ICP registration of two sets of corresponding point clouds (refer to Gao Xiang's "Fourteen Lectures on Visual Slam"), and the 3D Euclidean transformation matrix can be solved by singular value decomposition (SVD);

[0077] The calculated three-dimensional Euclidean transformation matrix is ​​applied to all the scaled arrow template points to obtain arrow annotations that fit the three-dimensional point cloud of the road surface, thereby determining the size and direction information of the actual road surface arrows.

[0078] The embodiments of the present invention are given for the purpose of illustration and description, and are not intended to be exhaustive or to limit the invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

Claims

1. A method for quickly marking arrows on a road point cloud, characterized in that: The method for quickly marking arrows on a road surface point cloud comprises the following steps: S1. Point cloud data collection: Scanning the actual road surface using a 3D laser scanning system to generate original point cloud data containing road surface information; S2. Annotate key points on the point cloud: Identify and extract key points of road arrows from the original point cloud data through the point cloud data annotation platform; S3. Calculate the transformation relationship: Compare the three key points in the original point cloud data generated by the actual road surface with the three key points corresponding to the template to obtain the corresponding scaling, rotation, and translation transformation parameters; S4. Application of transformation parameters: The key points in the template apply the transformation parameters to obtain the positions of all points including the key points indicated by arrows in the original point cloud data generated by the actual road surface; In step S3, the transformation of parameters includes the following operations: The three key points in the point cloud of the actual road arrow are A, B, and C, and the key points of the arrow template are A_t, B_t, and C_t; Then calculate the scaling ratio and find the distance between any two of the three points: Dis_AB,Dis_BC,Dist_AC,Dis_AB_t,Dis_BC_t,Dis_AC_t; Then calculate the distance ratio: ScaleRatio_AB=Dis_AB / Dis_AB_t,ScaleRatio_BC=Dis_BC / Dis_BC_t,ScaleR atio_AC=Dis_AC / Dis_AC_t; After obtaining the three distance ratios, calculate the average of the three distance ratios to obtain the same scaling ratio for all template key points: ScaleRatio_avg=(ScaleRatio_AB+ScaleRatio_BC+ScaleRatio_AC) / 3; Then all arrow template key points are scaled by the ratio ScaleRatio_avg.

2. A method for quickly marking arrows on a road surface point cloud according to claim 1, characterized in that: The transformation parameters also include the following operations: The three-dimensional Euclidean transformation matrix from the scaled three arrow template points to the three point cloud key points of the actual road arrow is calculated, and the three-dimensional Euclidean transformation matrix is ​​solved by singular value decomposition.

3. The method for quickly marking arrows on a road surface point cloud according to claim 2, characterized in that: The transformation parameters also include the following operations: The calculated three-dimensional Euclidean transformation matrix is ​​applied to all the scaled arrow template points to obtain arrow annotations that fit the three-dimensional point cloud of the road surface, thereby determining the size and direction information of the actual road surface arrows.

4. The method for quickly marking arrows on a road surface point cloud according to claim 2, characterized in that: The arrow template is adapted to the actual road arrows.

5. The method for quickly marking arrows on a road surface point cloud according to claim 1, characterized in that: In step S1, the road surface information refers to road surface elements, which include but are not limited to arrows and lane lines.

6. A method for quickly marking arrows on a road surface point cloud according to any one of claims 1 to 5, characterized in that: The method for quickly marking arrows on a road surface point cloud is applied to the field of road surface arrow marking.

Images